Power dispatching method, power dispatching device, electronic device, and storage medium

Abstract

The embodiment of the application provides a kind of electric power scheduling method, electric power scheduling device, electronic equipment and storage medium, belong to electric power scheduling technical field.The method comprises: according to feasible interval and binary indicator variable, the approximate model of electric power scheduling model is determined, and reference scheduling model is obtained;Scene path is obtained by sampling scene tree, and the power generation state variable, binary indicator variable and local decision variable of each bus node in scene path are solved by reference scheduling model, and node solution value is obtained;According to node solution value, enhanced cut plane is generated;According to enhanced cut plane, the optimal solution of the power generation state variable of target bus node is calculated to obtain target power generation state, and the optimal solution of the local decision variable of target bus node is calculated to obtain target decision value;According to node solution value, target power generation state and target decision value, electric power network is scheduled.The embodiment of the application can improve the efficiency of electric power scheduling.

Description

Technical Field

[0001] This application relates to the field of power dispatching technology, and in particular to a power dispatching method, power dispatching device, electronic device and storage medium. Background Technology

[0002] In related technologies, power dispatching models are used to dispatch power networks. Power dispatching models typically employ multi-stage stochastic mixed integer programming models. The uncertainty of the model increases exponentially with the number of dispatching stages, and the variables to be solved include integers and continuous types, making the solution problem of the power dispatching model non-convex and non-continuous, increasing the difficulty of solving the model and resulting in low efficiency of power dispatching. Summary of the Invention

[0003] The main objective of this application is to provide a power dispatching method, power dispatching device, electronic device, and storage medium, which aim to improve the efficiency of power dispatching.

[0004] To achieve the above objectives, a first aspect of this application proposes a power dispatching method, the method comprising:

[0005] Obtain the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on a scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stages, and the bus nodes have generation state variables and local decision variables;

[0006] Determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein the binary indicator variable is used to indicate whether the power generation state variable is located within the feasible interval;

[0007] Based on the feasible interval and the binary indicator variable, an approximate model of the power dispatching model is determined to obtain a reference dispatching model;

[0008] The scenario tree is sampled to obtain the scenario path, and the generation state variable, the binary indicator variable and the local decision variable of each bus node in the scenario path are solved by the reference scheduling model to obtain the node solution value;

[0009] An enhanced cutting plane is generated based on the node solution values;

[0010] Based on the enhanced cut plane, the optimal solution of the power generation state variable of the target bus node is calculated to obtain the target power generation state, and the optimal solution of the local decision variable of the target bus node is calculated to obtain the target decision value; wherein, the target bus node is the root node of the scene tree;

[0011] Power dispatch is performed on the power network based on the node solution value, the target power generation state, and the target decision value.

[0012] In some embodiments, the step of solving for the generation state variables, the binary indicator variables, and the local decision variables of each bus node in the scenario path using the reference scheduling model to obtain the node solution value includes:

[0013] Traverse each bus node of the scene path forward to determine the parent node of the bus node;

[0014] The parent node's state value for the power generation state variable, the variable value and feasible interval for the binary indicator variable, and the approximate power generation cost and feasible interval of the bus node are substituted into the reference scheduling model to solve for the power generation state variable, the binary indicator variable, and the local decision variable of the bus node, thereby obtaining the node solution value of the bus node.

[0015] In some embodiments, generating the enhanced cut plane based on the nodal solution values ​​includes:

[0016] Starting from the second-to-last level, traverse each bus node of the scene path backward, refine the corresponding feasible interval based on the node solution value of the bus node, and determine the binary indicator variable of the refined feasible interval.

[0017] Determine the parent node of the bus node, and construct a Lagrange dual model based on the state value of the parent node for the power generation state variable, the variable value for the binary indicator variable, and the approximate power generation cost of the bus node;

[0018] Solving the Lagrange dual model yields the optimal dual multipliers and intercepts;

[0019] The enhanced cut plane is generated based on the optimal dual multiplier and the intercept.

[0020] In some embodiments, solving the Lagrange dual model to obtain the optimal dual multipliers and intercepts includes:

[0021] The first cutting plane is generated according to the Pareto optimal Lagrange cutting plane strategy;

[0022] Maximize the first cutting plane to solve the Lagrange dual model and obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model;

[0023] The intercept is calculated based on the optimal dual multiplier and the model output value.

[0024] In some embodiments, solving the Lagrange dual model to obtain the optimal dual multipliers and intercepts includes:

[0025] The second cutting plane is generated based on the square minimization cutting plane strategy;

[0026] The Lagrange dual model is solved by minimizing the second cutting plane to obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model;

[0027] The intercept is calculated based on the optimal dual multiplier and the model output value.

[0028] In some embodiments, the step of power dispatching of the power network based on the node solution value, the target generation state, and the target decision value includes:

[0029] Calculate the upper bound of the feasible region based on the node solution values;

[0030] The lower bound of the feasible region is calculated using the reference scheduling model based on the target power generation state and the target decision value;

[0031] Calculate the gap between the lower bound of the feasible region and the upper bound of the feasible region;

[0032] If the gap is less than or equal to a preset threshold, power dispatch is performed based on the target power generation state and the target decision value.

[0033] In some embodiments, calculating the upper bound of the feasible region based on the node solution value includes:

[0034] Calculate the path cost of the scene path based on the node solution value;

[0035] Calculate the mean and standard deviation of the path cost;

[0036] The upper bound of the feasible region is calculated based on the mean and the standard deviation.

[0037] To achieve the above objectives, a second aspect of this application provides a power dispatching device, the device comprising:

[0038] The acquisition module is used to acquire the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on a scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stages, and the bus nodes have generation state variables and local decision variables;

[0039] The first determining module is used to determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein, the binary indicator variable is used to indicate whether the power generation state variable is located within the feasible interval;

[0040] The second determining module is used to determine an approximate model of the power dispatching model based on the feasible interval and the binary indicator variable, so as to obtain a reference dispatching model;

[0041] The sampling module is used to sample the scenario tree to obtain the scenario path, and solve the generation state variable, the binary indicator variable and the local decision variable of each bus node in the scenario path through the reference scheduling model to obtain the node solution value;

[0042] The generation module is used to generate an enhanced cutting plane based on the node solution values;

[0043] The calculation module is used to calculate the optimal solution of the power generation state variable of the target bus node to obtain the target power generation state based on the enhanced cut plane, and to calculate the optimal solution of the local decision variable of the target bus node to obtain the target decision value; wherein, the target bus node is the root node of the scene tree;

[0044] The scheduling module is used to perform power scheduling on the power network based on the target power generation status and the target decision value.

[0045] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0046] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0047] The power dispatching method, power dispatching device, electronic device, and computer-readable storage medium proposed in this application obtain a power dispatching model of a power network. The power dispatching model is a multi-stage stochastic mixed integer programming model. To reduce the difficulty of solving the model, the power dispatching model is redefined on a scenario tree. The scenario tree includes bus nodes connected sequentially according to the dispatching stages. Each bus node has generation state variables and local decision variables. The generation state variables are of mixed integer type, and the local decision variables include continuous variables and integer variables. To solve for each variable, the feasible interval of the generation state variables and the binary indicator variable of the feasible interval are determined, so as to improve the state space through the binary indicator variable. An approximate model of the power dispatching model is determined based on the feasible interval and the binary indicator variable, so as to gradually approximate the accurate representation of the non-convex value function and obtain a reference dispatching model. Path sampling is performed on the scenario tree to obtain scenario paths, and the generation state variables, binary indicator variables, and local decision variables of each bus node in the scenario path are solved through the reference dispatching model to obtain node solution values. An enhanced cut plane is generated based on the node solution values. By generating an enhanced cut plane in the lift space, the algorithm is ensured to converge to the global optimum of the multi-stage stochastic mixed-integer programming model with a certain probability, solving the problem that traditional methods can only obtain local optima due to non-convexity. Based on the enhanced cut plane, the optimal solution of the generation state variables of the target bus node is calculated to obtain the target generation state, and the optimal solution of the local decision variables of the target bus node is calculated to obtain the target decision value, thus obtaining the global optimum and reducing the difficulty of model solving. Power dispatching is performed on the power network based on the node solution values, the target generation state, and the target decision value, improving both the efficiency and accuracy of power dispatching. Attached Figure Description

[0048] Figure 1 This is a flowchart of the power dispatching method provided in the embodiments of this application;

[0049] Figure 2 yes Figure 1 The flowchart of step S140 in the middle;

[0050] Figure 3 yes Figure 1 The flowchart of step S150 in the middle;

[0051] Figure 4 yes Figure 3 The flowchart of step S330 in the process;

[0052] Figure 5 yes Figure 3 Another flowchart of step S330 in the process;

[0053] Figure 6 yes Figure 1 The flowchart of step S170 in the process;

[0054] Figure 7 yes Figure 6 The flowchart of step S610 in the text;

[0055] Figure 8 This is a schematic diagram of the structure of the power dispatching device provided in the embodiments of this application;

[0056] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0058] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0060] Multistage Stochastic Mixed-Integer Programming (MS-SMIP) is a mathematical optimization framework for modeling sequential decision problems, widely used in power system planning, logistics scheduling, and medical resource allocation. Related technologies construct power dispatching models as multistage stochastic mixed-integer programming models and utilize these models for power network dispatching. However, the challenge of MS-SMIP lies in the fact that its uncertainty increases exponentially with the number of dispatching stages, and the variables to be solved include both integer and continuous types, resulting in a non-convex and discontinuous problem that is difficult to solve, ultimately leading to low efficiency in power dispatching.

[0061] Based on this, embodiments of this application provide a power dispatching method, a power dispatching device, an electronic device, and a computer-readable storage medium, aiming to improve the efficiency of power dispatching.

[0062] The power dispatching method, power dispatching device, electronic device, and computer-readable storage medium provided in the embodiments of this application are specifically described through the following embodiments. First, the power dispatching method in the embodiments of this application is described.

[0063] The power dispatching method provided in this application relates to the field of power dispatching technology. The power dispatching method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the power dispatching method, but is not limited to the above forms.

[0064] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0065] Figure 1 This is an optional flowchart of the power dispatching method provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S110 to S170.

[0066] Step S110: Obtain the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on the scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stage, and the bus nodes have generation state variables and local decision variables;

[0067] Step S120: Determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein, the binary indicator variable is used to indicate whether the power generation state variable is located in the feasible interval.

[0068] Step S130: Determine an approximate model of the power dispatching model based on the feasible interval and the binary indicator variables to obtain the reference dispatching model;

[0069] Step S140: Path sampling is performed on the scenario tree to obtain the scenario path, and the generation state variables, binary indicator variables and local decision variables of each bus node in the scenario path are solved by the reference scheduling model to obtain the node solution value.

[0070] Step S150: Generate an enhanced cut plane based on the nodal solution values;

[0071] Step S160: Based on the enhanced cut plane, calculate the optimal solution of the power generation state variables of the target bus node to obtain the target power generation state, and calculate the optimal solution of the local decision variables of the target bus node to obtain the target decision value; wherein, the target bus node is the root node of the scene tree;

[0072] Step S170: Perform power dispatch on the power network based on the node solution value, the target power generation state, and the target decision value.

[0073] In step S110 of some embodiments, a power dispatch model of the power network is obtained. The power dispatch model is a multi-stage stochastic mixed integer programming model with T dispatch stages. The model is displayed in the scenario tree. The scenario tree is defined as follows: It has multiple nodes, each with state variables and local decision variables. State variables describe the network state of the power network and are of mixed integer type. Local decision variables provide control instructions for the power network, connecting consecutive scheduling phases and influencing its network state. Local decision variables include both continuous and integer variables. The depth of the scenario tree is the same as the number of scheduling phases; one level of the scenario tree represents one scheduling phase. Each node... It has a unique parent node Arrive at node The probability is The root node is node 1, and the initial condition is... , .in, This represents the state variable of the parent node of the root node. This represents the probability of reaching the root node. It should be noted that since the root node has no parent node, therefore... The value is 0. The value is 1.

[0074] The power dispatch model aims to minimize the expected total cost. The model comprises multiple dispatch phases, and uncertainty is described by a scenario tree. Each node's decision is influenced by the state of its parent node and the current stochastic implementation. The power dispatch model defined by the scenario tree can be represented as:

[0075] ,

[0076] ,

[0077] ,

[0078] in, Represents the total value function; A collection of nodes representing the scene tree; This represents the state variable of the nth node, which includes continuous and integer components; This represents the local decision variable of the nth node; and These represent the cost coefficients of the state variables and the cost coefficients of the local decision variables, respectively. , , , It can be a constraint matrix or a constraint vector. and It defines the feasible region of variable boundaries and other linear or integer constraints.

[0079] State variables are represented as vectors, using subscripts. Represents state variables The Each component, denoted as For the root node, the recursive value function is defined as:

[0080] ,

[0081] ,

[0082] ,

[0083] in, The value function representing the root node.

[0084] For non-root nodes, the recursive value function is defined as:

[0085] ,

[0086] ,

[0087] ,

[0088] in, The value function representing the non-root node n; For nodes The set of child nodes; Indicates from node To its child nodes The conditional probability; Represents a node The value function at the location; superscript This indicates the transpose operation.

[0089] It should be noted that for any leaf node Since leaf nodes do not have a future scheduling phase, they can be set... And define the future cost term in its value function as 0, the future cost term is the third term of the value function.

[0090] It should be further noted that, for the above model, all coefficients , , , , , , All are rational numbers. (Set) and It is a rational and compact mixed-integer linear program that can represent the set. State variables. Each component Each has clearly defined upper and lower boundaries. , express The index set has , , for A subset of represents the range of values ​​for the j-th component. The Cartesian product of all component value intervals. The model has relatively complete recourse, i.e. , yes The domain of definition. The value function of each node is lower-bounded, i.e. , , This represents the lower bound of the value function of node n.

[0091] Power networks include busbars ,load and generator Connected to the busbar A subset of the load is represented as Connected to the busbar A subset of the generators is represented as In the scenario tree, nodes are bus nodes in the power network. The scenario tree includes bus nodes connected sequentially according to the order of scheduling phases. Each bus node has generation state variables and local decision variables. Generation state variables describe the electrical state of the bus node and can be generator switching status, generation level, power flow, etc. Local decision variables control the electrical state of the bus and can be load attenuation ratios.

[0092] Consider the multi-period stochastic unit commitment problem with binary and continuous state variables, the bus node of the scenario tree The power generation state variables of the bus node include binary indicator variables used to represent the generator start-up state. Power-on operation Power off action and the continuous power generation level of each generator connected to the bus node. Through local decision variables This indicates the percentage of load reduction. Total cost includes the cost of generating electricity. Startup costs Shutdown costs and the penalty rate The cost of load reduction. The cost of generating power is represented by a piecewise linear function, where each segment... It is determined by the intercept and slope A parameterized linear function. The total value function in the power dispatch model is expressed as:

[0093] ,

[0094] in, This represents the load retention ratio of the load d connected to bus node n.

[0095] Output cost constraint is expressed as:

[0096] ,

[0097] in, For connecting bus nodes generator The cost of generating electricity.

[0098] DC power flow constraints are expressed as:

[0099] ,

[0100] in, Indicates time branch road The active power between bus node i and bus node j; This represents the susceptance between node i and node j; This represents the voltage phase angle at bus node i at time t; This represents the voltage phase angle at bus node j at time t.

[0101] The boundary constraints for power flow on a transmission line are expressed as follows:

[0102] ,

[0103] in, and These represent the lower and upper power bounds between node i and node j, respectively.

[0104] The boundary constraints for the power generation level on the transmission line are expressed as follows:

[0105] ,

[0106] in, and These represent the connecting bus nodes. generator The lower and upper limits of power generation levels.

[0107] The power flow balance constraint is expressed as:

[0108] ,

[0109] in, This represents the load demand power of the load d connected to bus node n; This represents the load retention ratio of the load d connected to bus node n.

[0110] The generator start-up and shutdown constraints are expressed as follows:

[0111] ,

[0112] This constraint links the on / off state to the on / off decision.

[0113] Minimum operating time and minimum downtime constraints are enforced on the generator. The minimum operating time constraint is expressed as:

[0114] ,

[0115] Where UT represents the minimum running time; set Includes slave nodes The Ancestors to nodes Each node in the path between them.

[0116] The minimum downtime constraint is expressed as:

[0117] ,

[0118] Where UD represents the minimum downtime.

[0119] A ramp constraint is enforced on the generator, and the ramp constraint is expressed as follows:

[0120] ,

[0121] ,

[0122] in, This represents the climbing boundary parameters.

[0123] Other constraints are expressed as follows:

[0124] ,

[0125] ,

[0126] in, , , All are binary variables, and their values ​​are either 0 or 1; The value of is within the range of [0,1].

[0127] In step S120 of some embodiments, algorithm parameters are initialized. These parameters include the value of the iteration counter, the upper bound of the feasible region, the lower bound of the feasible region, partition information, and an approximation of the future value function for all bus nodes. The iteration counter is initialized to 1, the lower bound of the feasible region is set to negative infinity, and the upper bound of the feasible region is set to positive infinity. Simultaneously, the number of sampling paths M and the maximum number of iterations are set. Tolerance parameters Parameters such as... are defined. An initial partition is defined for each state component of the power generation state variable based on the lower and upper bounds of the feasible region, resulting in a feasible interval. A binary indicator variable for the feasible interval is then initialized. This binary indicator variable is used to indicate whether a state component of the power generation state variable lies within the feasible interval. For each node... Initialize its approximate future value function For a lower bound To reduce the difficulty of solving power dispatching models, this application's embodiments introduce a binary indicator variable based on a state-space lifting mechanism to dynamically partition and lift the state space, thereby achieving an accurate approximation of non-convex functions.

[0128] Initialize the state space as The initial state space is divided, and each state component of each bus node is assigned an interval to obtain the feasible interval of the state component. For each bus node... The j-th state component, Its feasible interval is , and Representing the lower bound and upper bound of the interval respectively, the index set For bus node n, the initial partition of the entire state space is denoted as follows: .

[0129] When initializing the binary indicator variable, a binary indicator variable is introduced for each feasible interval. ,in , Indicates the index of the feasible interval. Defines the set of binary indicator variables for the generation state variables of bus node n. for:

[0130] ,

[0131] in, This represents the set of binary indicator variables corresponding to bus node n; This represents the j-th state component in the power generation state variables of bus node n. Two binary indicator variables are used to indicate state components. Is it within the feasible range? Divided intervals Inside; and Representing intervals The lower bound and upper bound of the interval.

[0132] In step S130 of some embodiments, the power dispatch model is redefined based on the feasible interval and the binary indicator variable to obtain an approximate model of the power dispatch model, thus obtaining a reference dispatch model. For non-root nodes, the reference dispatch model is expressed as:

[0133] ,

[0134] ,

[0135] ,

[0136] ,

[0137] in, This represents the approximate feasible interval corresponding to bus node n; This represents an approximation of the value function of bus node n; This represents an approximation of the future value function; and All are auxiliary variables.

[0138] auxiliary variables Equivalent to the state of the parent node auxiliary variables Equivalent to the parent node's binary indicator variable, this is often referred to as an unexpected constraint.

[0139] For the root node, the reference scheduling model is represented as follows:

[0140] ,

[0141] ,

[0142] ,

[0143] in, This represents an approximation of the value function of the root node; This represents the approximate feasible interval of the root node; This represents an approximation of the future value function.

[0144] Approximate cuts of the future value function Represented as:

[0145] ,

[0146] in, , and Indicates the first Each weighting coefficient; This indicates the number of weighting coefficients.

[0147] Please see Figure 2 In some embodiments, step S140 may include, but is not limited to, steps S210 to S220:

[0148] Step S210: Traverse each bus node of the scene path forward to determine the parent node of the bus node;

[0149] Step S220: Substitute the state value of the parent node for the power generation state variable, the variable value and feasible interval for the binary indicator variable, and the approximate power generation cost and feasible interval of the bus node into the reference scheduling model to solve for the power generation state variable, binary indicator variable and local decision variable of the bus node, and obtain the node solution value of the bus node.

[0150] In step S210 of some embodiments, in the first In each iteration, starting from the root node of the scene tree, M scene paths are generated independently. For each scene path The algorithm sequentially solves the forward problem of each node along the scenario path from the first scheduling stage (root node) to the last scheduling stage (leaf node). By providing the state solution of the parent node and an approximation of the current node's value function, the state solution and decision solution of the current node can be obtained. The state solution represents the solution for the power generation state variables, and the decision solution represents the solution for the local decision variables. During the solution process, each bus node along the scenario path is traversed forward from the root node to the leaf node, and the parent node of each bus node is determined. It should be noted that if a bus node is the root node, its parent node is null.

[0151] In step S220 of some embodiments, the approximate generation cost is an approximation of the future value function. For the scenario path... bus node The parent node's state value for the power generation state variable For the values ​​of binary indicator variables and feasible interval Approximate power generation cost of bus node n and feasible interval Substituting the reference scheduling model to solve the approximate problem in the lift space ( Solve the first... The optimal state value of the generator state variables of the bus node in the next iteration. Optimal variable values ​​of binary indicator variables and the optimal decision value of local decision variables This yields the node solution values ​​for the bus nodes.

[0152] The node solution value of the current bus node is passed to its child nodes as input for solving the approximate problem of the child nodes, until the node solution value of each bus node is obtained.

[0153] Through the above steps S210 to S220, the optimal solution for each node can be calculated, and power dispatch can be performed based on the optimal solution to improve the efficiency of power dispatch.

[0154] Please see Figure 3 In some embodiments, step S150 may include, but is not limited to, steps S310 to S340:

[0155] Step S310: Starting from the second to last level, traverse each bus node of the scene path backward, refine the corresponding feasible interval based on the node solution value of the bus node, and determine the binary indicator variable of the refined feasible interval.

[0156] Step S320: Determine the parent node of the bus node, and construct a Lagrange dual model based on the parent node's state value for the power generation state variable, the variable value for the binary indicator variable, and the approximate power generation cost of the bus node.

[0157] Step S330: Solve the Lagrange dual model to obtain the optimal dual multipliers and intercepts;

[0158] Step S340: Generate the enhanced cut plane based on the optimal dual multiplier and intercept.

[0159] In step S310 of some embodiments, for each scene path, starting from the second-to-last level, i.e., stage T-1, the path is traversed backwards to the root node, i.e., stage 1. The parent-child relationships between nodes in the forward and backward traversals are reversed. For each bus node of the scene path, the node solution value of the bus node obtained from the forward sampling is used... That is, improve the state solution for the corresponding feasible interval. The feasible interval is further refined by dividing it into two sub-intervals, resulting in the refined feasible interval. Specifically, for the state components of the power generation state variable... Get the active range where its current state value is located. The active region is divided into two sub-regions for partitioning refinement. The refinement strategy can employ either the current solution partitioning strategy or an equal-division partitioning strategy. The refinement strategy supports dynamic adjustment of the partitions based on the current solution or a bisection method to balance approximate accuracy and computational burden. The active region is defined as satisfying... feasible range .

[0160] The current solution partitioning strategy involves the following partitioning process: using the current state as the solution... As the dividing point, forming and Two new zones.

[0161] The segmentation process of the equal division strategy is as follows: using the midpoint of the interval... As the dividing point, forming and Two new zones.

[0162] Update partitioned index set And introduce new binary indicator variables for the newly generated intervals. The calculation of the binary indicator variable can be referred to step S120, and will not be repeated here.

[0163] In step S320 of some embodiments, in order to generate the Lagrange cut plane, it is also necessary to update the Lagrange dual problem in the lifting space. Determine the bus node n. Based on the state value of the parent node for the power generation state variable For the values ​​of binary indicator variables Approximate power generation cost of bus node We construct a sparse lifting Lagrange dual model. The Lagrange dual model is expressed as:

[0164] ,

[0165] ,

[0166] ,

[0167] ,

[0168] ,

[0169] ,

[0170] ,

[0171] in, Representing the Lagrange duality problem; Let represent the objective function of the Lagrange dual model; Auxiliary variables representing the parent node; This represents the approximate lower bound of the future value function.

[0172] like express The index of the active element, for each Then the current .

[0173] In step S330 of some embodiments, a sparse technique is used to introduce dual multipliers into the binary indicator variables corresponding to the currently active partition elements, thereby significantly reducing the search dimensionality of the dual multipliers. The Lagrange dual model is then solved to obtain the optimal dual multipliers and intercepts.

[0174] In step S340 of some embodiments, after solving the dual problem, a Lagrange cut plane is generated based on the optimal dual multipliers and intercepts, resulting in an enhanced cut plane. When generating the Lagrange cut plane of the boost space, a sparsity technique is employed, introducing non-zero dual multipliers only for the currently active partition elements. This significantly reduces the dimensionality of the Lagrange dual problem, alleviates the computational burden, and makes the boosting process more computationally feasible, especially suitable for large-scale power dispatching problems. The enhanced cut plane employs a sparse cut plane technique, which can significantly boost the lower bound within a few iterations, improving computational efficiency and thus improving the efficiency of power dispatching. The Lagrange cut plane is represented as:

[0175] ,

[0176] in, Indicates the intercept; This represents the optimal dual multiplier.

[0177] This enhanced cut plane can be added to the node. Approximation of the future value function In, thus updating .

[0178] Steps S310 to S340 above improve the state space by dynamically introducing binary indicator variables, raising the state space to a high-dimensional space containing the original state variables and binary indicator variables representing the elements of the current partition. An enhanced cutting plane is generated in the improved space, enabling the algorithm to gradually approximate the accurate representation of the non-convex value function. Theoretically, this guarantees that the algorithm can converge to the global optimal solution of the multi-stage stochastic mixed integer programming with a certain probability. This solves the problem that traditional methods can only obtain local optima or relaxed lower bounds due to non-convexity, and improves the accuracy of power dispatching.

[0179] To obtain high-quality cutting planes, this application employs a Pareto optimal Lagrange cut plane strategy (PLC) and a squared minimum cut plane strategy (SMC) to generate cutting planes, thereby improving the geometric strength and convergence efficiency of the cutting planes. The Pareto optimal Lagrange cut plane strategy, while ensuring the compactness of the convex hull of the value function at a given point, generates the cutting plane by maximizing the value at a core point (usually a point inside the partitioned convex hull), thus achieving a stronger global approximation. The squared minimum cut plane strategy, while maintaining the compactness of the cutting plane, generates it by minimizing the L2 norm of the dual multipliers, resulting in a flatter cutting plane and improving the approximation at other points. This application employs these two enhancement techniques—Pareto optimal Lagrange cut plane and squared minimum cut plane—to optimize the quality of the cutting planes through geometric properties. These two enhancement techniques significantly accelerate the convergence process of the algorithm and improve computational efficiency. The embodiments of this application combine lifting space and enhanced cutting plane to accurately characterize non-convex functions, so that the algorithm converges to the global optimum rather than a local approximation, thereby ensuring global convergence for multi-stage stochastic mixed integer programming problems.

[0180] Please see Figure 4 In some embodiments, step S330 may include, but is not limited to, steps S410 to S430:

[0181] Step S410: Generate the first cutting plane according to the Pareto optimal Lagrange cutting plane strategy;

[0182] Step S420: Maximize the first cutting plane to solve the Lagrange dual model and obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model;

[0183] Step S430: Calculate the intercept based on the optimal dual multipliers and the model output value.

[0184] In step S410 of some embodiments, a first cutting plane is generated according to a Pareto-optimal Lagrange cutting plane strategy. Specifically, a core point is selected. , Represents the state variables of power generation. This represents a binary indicator variable. The point is located relatively inside the convex hull of the state space of the parent node of bus node n. A simple and efficient choice is to take the combination of the midpoints of each feasible interval of each power generation state variable and its corresponding binary indicator variable.

[0185] In step S420 of some embodiments, the first cutting plane is represented as:

[0186] ,

[0187] ,

[0188] in, The optimal value is obtained by solving the Lagrange duality problem. It is a small tolerance parameter.

[0189] Maximizing the first cutting plane to solve the Lagrange dual model yields the optimal model value, which includes the optimal dual multiplier. Model output values ​​of the Lagrange dual model The optimal dual multiplier is the enhanced dual multiplier.

[0190] In step S430 of some embodiments, the intercept is calculated based on the optimal dual multipliers and the model output value. The formula for calculating the intercept is as follows:

[0191] ,

[0192] in, This represents the intercept.

[0193] Through the above steps S410 to S430, the optimal dual multiplier and intercept can be obtained, so as to generate an enhanced cut plane that is maximized at the core point based on the optimal dual multiplier and intercept.

[0194] Please see Figure 5 In some embodiments, step 330 may include, but is not limited to, steps S510 to S530:

[0195] Step S510: Generate the second cutting plane according to the square minimization cutting plane strategy;

[0196] Step S520: Minimize the second cutting plane to solve the Lagrange dual model and obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model;

[0197] Step S530: Calculate the intercept based on the optimal dual multipliers and the model output value.

[0198] In step S510 of some embodiments, a second cutting plane is generated according to a square minimization cutting plane strategy. The second cutting plane is defined as follows: , This indicates the dual multiplier.

[0199] In step S520 of some embodiments, the second cleaving plane is minimized to solve the Lagrange dual model and obtain the optimal model value, which includes the optimal dual multiplier and the model output value of the Lagrange dual model. That is:

[0200] ,

[0201] ,

[0202] in, This represents the model's output value; This represents the optimal dual multiplier.

[0203] In step S530 of some embodiments, the intercept is calculated based on the optimal dual multipliers and the model output value. The formula for calculating the intercept is as follows:

[0204] ,

[0205] in, This represents the intercept.

[0206] Through the above steps S510 to S530, the optimal dual multiplier and intercept can be obtained, so as to generate a flatter and more numerically stable enhanced cutting plane based on the optimal dual multiplier and intercept.

[0207] In step S160 of some embodiments, an enhanced cut plane is added to the approximation of the future value function of the reference scheduling model, and the forward problem of the root node is solved through the reference scheduling model. The target power generation state is obtained by determining the optimal solution of the power generation state variables of the target bus node, and the target decision value is obtained by determining the optimal solution of the local decision variables of the target bus node. The target bus node is the root node of the scenario tree.

[0208] Please see Figure 6In some embodiments, step 170 may include, but is not limited to, steps S610 to S640:

[0209] Step S610: Calculate the upper bound of the feasible region based on the node solution values;

[0210] Step S620: Calculate the lower bound of the feasible region based on the target generation state and the target decision value using the reference scheduling model;

[0211] Step S630: Calculate the gap between the lower bound and the upper bound of the feasible region;

[0212] In step S640, if the gap is less than or equal to a preset threshold, power dispatch is performed based on the target power generation status and the target decision value.

[0213] In step S610 of some embodiments, the upper bound of the feasible region is calculated based on the node solution values ​​of the bus nodes on each scenario path to update the initial upper bound of the feasible region.

[0214] In step S620 of some embodiments, the target power generation state and the target decision value are substituted into the reference scheduling model to calculate the output value of the reference scheduling model and obtain the optimal value. This optimal value is taken as the global lower bound for the current iteration, i.e. To update the lower bound of the feasible domain.

[0215] In step S630 of some embodiments, the gap between the lower bound and the upper bound of the feasible region is calculated, and the calculation formula is expressed as:

[0216] ,

[0217] Where Gap represents the gap; UB represents the upper bound of the feasible region; and LB represents the lower bound of the feasible region. This represents the absolute value of the lower bound of the feasible region.

[0218] In step S640 of some embodiments, if the gap is less than or equal to a preset threshold, i.e. , This indicates a preset threshold, or that the current iteration count has reached the maximum iteration count. If the computation time limit is reached, the algorithm terminates, and the target generation state and target decision value are used as the optimal decision strategy for the root node to perform power dispatch based on the target generation state and target decision value. Otherwise, the iteration counter is updated, and the algorithm is set to... The feasible range can be carried with the updated partition information. Approximation of the future value function Return to step S140 and continue execution.

[0219] Through the above steps S610 to S640, the global optimal value can be obtained, and power dispatch can be performed based on the global optimal value to improve the accuracy of power dispatch.

[0220] Please see Figure 7 In some embodiments, step S610 may include, but is not limited to, steps S710 to S730:

[0221] Step S710: Calculate the path cost of the scene path based on the node solution value;

[0222] Step S720: Calculate the mean and standard deviation of the path cost;

[0223] Step S730: Calculate the upper bound of the feasible region based on the mean and standard deviation.

[0224] In step S710 of some embodiments, the node solution value includes the state value of the bus node for the generation state variable and the decision value for the local decision variable. For each scenario path, the cost of the bus node is calculated based on the state value and decision value of the bus node in the scenario path. The costs of all bus nodes from the root node to the leaf node are added together, and the statistically expected cost is calculated to obtain the path cost of the scenario path. The formula for calculating the path cost is expressed as:

[0225] ,

[0226] in, Representing scene path Path cost; Representing scene path The set of bus nodes included; Indicates the first The next iteration; and Indicates the cost coefficient; and These represent the power generation state variables and local decision variables, respectively.

[0227] In step S720 of some embodiments, the mean and standard deviation of the path costs of the M scene paths are calculated, and the formulas for calculating the mean and standard deviation are expressed as follows:

[0228] ,

[0229] ,

[0230] in, This represents the mean; Indicates variance; It represents the standard deviation.

[0231] In step S730 of some embodiments, the upper bound of the feasible region is calculated based on the mean and standard deviation to update the initial upper bound of the feasible region. The formula for calculating the upper bound of the feasible region is expressed as:

[0232] ,

[0233] Where UB represents the upper bound of the feasible region; and These represent the mean and standard deviation, respectively; M represents the number of scene paths. Represents the standard normal distribution Quantiles.

[0234] Through steps S710 to S730, the upper bound of the feasible region can be updated, thereby continuously converging the upper bound of the feasible region and improving the accuracy of power dispatch.

[0235] The power dispatching method of this application has broad applicability and can handle general mixed-integer state variables without relying on special problem structures. The embodiments of this application demonstrate the enhanced performance of the proposed power dispatching method in multi-stage stochastic mixed-integer programming problems through numerical experiments. Numerical tests were conducted on the IEEE 30-bus test case, which covers 6, 8, or 12 cycles, each cycle corresponding to one stage. Uncertain demand sequences... The modeling is as follows: Create a nominal demand sequence where the values ​​for the first six stages are equal to 90% of the static demand, and the values ​​for the seventh and subsequent stages are equal to 130% of the static demand. Additionally, each stage is appended with a uniformly distributed [value / value]. The random fluctuation factor was assumed. It was assumed that the stages were independent of each other, with 5 or 10 scenarios per stage. All optimization models were built using JuMP in Julia v1.11 and solved using Gurobi. Experiments were conducted on a workstation equipped with an M3 MAX CPU and 48 GB of memory, using 5 threads to execute the numerical test task concurrently. , Run all test instances for 3600 seconds, recording the upper and lower bounds, and calculating the time and number of iterations when the relative difference between the upper and lower bounds reaches 1%. For PLCs, for all... The core point is fixed as The Lagrange duality problem The solution is obtained using the level method. The convergence tolerance of the level method is set to... The maximum number of iterations is set to In the experiment, using to replace Therefore, it is not necessary to solve the Lagrange duality problem to obtain... .because Only provided The lower approximation may encounter infeasibility. When such infeasibility is encountered, the level method is terminated and the cutting coefficient from the previous iteration is output. This technique produces an efficient Lagrange cut, which follows the same reasoning when encountering numerical problems.

[0236] This application compares the computational performance of SDDP-L, SDDiP, and SDDP on the MSUC problem in Tables 1 to 4, and tests three different Lagrange cut types: LC, PLC, and SMC. The performance of two partitioning methods is also tested. Each stage has R scenarios, recording the optimal lower bound (LB) and the optimality gap (Gap) between the statistical upper and lower bounds. The tables also include the number of iterations within the time limit (Iter), the average time per iteration (Time / Iter.), the standard deviation (in parentheses), and the time and number of iterations required to reach the 1% optimality gap (Time1% and Iter1%).

[0237] The computational performance of the SDDP-L algorithm for the MSUC problem based on the current partitioning refinement is shown in Table 1:

[0238] Table 1

[0239]

[0240] The computational performance of the SDDP-L algorithm, which refines the partitions using a binary search method, for the MSUC problem is shown in Table 2.

[0241] Table 2

[0242]

[0243] The computational performance of the SDDiP algorithm in the MSUC problem is shown in Table 3:

[0244] Table 3

[0245]

[0246] The computational performance of the SDDP algorithm using Lagrange cut in the MSUC problem is shown in Table 4:

[0247] Table 4

[0248]

[0249] As can be seen from the table, for problems with complex non-convex characteristics like MSUC, even within the time limit, LC failed to provide a meaningful improvement in the lower bound, while PLC and SMC rapidly improved the lower bound, reducing the optimality gap to a 1% tolerance in most cases. This demonstrates that the embodiments of this application are well-suited for complex non-convex problems on a real-world industrial scale. In all test instances, SMC converged stably, and in instances with T=12 and R=10, the method achieved a 0.5% gap within a reasonable time, further verifying the robustness and reliability of the square minimization cut plane in complex environments. PLC exhibited a longer single-iteration time on the MSUC problem and failed to achieve a 1% gap in the largest instance, indicating that the Pareto optimal cut plane may be computationally burdensome and affect its practicality in extremely high-dimensional problems. Comparing different partitioning strategies, it was found that the partitioning strategy based on the current solution is superior to the bisection method in most cases, especially when using the enhanced cut plane. This indicates that dynamic partitioning based on the current solution can more effectively guide the refinement of the state space and form better synergy with the enhanced cut plane.

[0250] This application's embodiments combine lifting, enhanced cutting plane, and sparsification techniques to form a complete, efficient, and robust solver. Numerical experiments show that, compared to existing methods, this application's embodiments can obtain tighter lower bounds faster and converge to the global optimal solution that meets accuracy requirements with higher reliability in practical large-scale problems such as power generation expansion planning and stochastic unit combination.

[0251] Please see Figure 8 This application also provides a power dispatching device that can implement the above-described power dispatching method. The power dispatching device includes:

[0252] The acquisition module 810 is used to acquire the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on the scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stage, and the bus nodes have generation state variables and local decision variables;

[0253] The first determining module 820 is used to determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein, the binary indicator variable is used to indicate whether the power generation state variable is located in the feasible interval;

[0254] The second determining module 830 is used to determine an approximate model of the power dispatching model based on the feasible interval and the binary indicator variables, and obtain a reference dispatching model;

[0255] The sampling module 840 is used to sample the scenario tree to obtain the scenario path, and solve the generation state variables, binary indicator variables and local decision variables of each bus node in the scenario path through the reference scheduling model to obtain the node solution value.

[0256] Generation module 850 is used to generate enhanced cut planes based on nodal solution values;

[0257] The calculation module 860 is used to calculate the optimal solution of the power generation state variables of the target bus node to obtain the target power generation state based on the enhanced cut plane, and to calculate the optimal solution of the local decision variables of the target bus node to obtain the target decision value; wherein, the target bus node is the root node of the scene tree;

[0258] The scheduling module 870 is used to perform power dispatching on the power network based on the target power generation status and target decision values.

[0259] The specific implementation of this power dispatching device is basically the same as the specific implementation of the power dispatching method described above, and will not be repeated here.

[0260] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described power dispatching method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0261] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0262] The processor 910 can be implemented using a general-purpose central processing unit (CPU), microprocessor, application specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0263] The memory 920 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 920 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 920 and is called and executed by the processor 910 using the power dispatching method of the embodiments of this application.

[0264] The input / output interface 930 is used to implement information input and output;

[0265] The communication interface 940 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0266] Bus 950 transmits information between various components of the device (e.g., processor 910, memory 920, input / output interface 930, and communication interface 940);

[0267] The processor 910, memory 920, input / output interface 930 and communication interface 940 are connected to each other within the device via bus 950.

[0268] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described power dispatching method.

[0269] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0270] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0271] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0272] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0273] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0274] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0275] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0276] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0277] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0278] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0279] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0280] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A power dispatching method, characterized in that, The method includes: Obtain the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on a scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stages, and the bus nodes have generation state variables and local decision variables; Determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein the binary indicator variable is used to indicate whether the power generation state variable is located within the feasible interval; Based on the feasible interval and the binary indicator variable, an approximate model of the power dispatching model is determined to obtain a reference dispatching model; The scenario tree is sampled to obtain the scenario path, and the generation state variable, the binary indicator variable and the local decision variable of each bus node in the scenario path are solved by the reference scheduling model to obtain the node solution value; An enhanced cutting plane is generated based on the node solution values; Based on the enhanced cut plane, the optimal solution of the power generation state variable of the target bus node is calculated to obtain the target power generation state, and the optimal solution of the local decision variable of the target bus node is calculated to obtain the target decision value; wherein, the target bus node is the root node of the scene tree; Power dispatch is performed on the power network based on the node solution value, the target power generation state, and the target decision value.

2. The method according to claim 1, characterized in that, The step of solving for the generation state variables, the binary indicator variables, and the local decision variables of each bus node in the scenario path using the reference scheduling model to obtain node solution values ​​includes: Traverse each bus node of the scene path forward to determine the parent node of the bus node; The parent node's state value for the power generation state variable, the variable value and feasible interval for the binary indicator variable, and the approximate power generation cost and feasible interval of the bus node are substituted into the reference scheduling model to solve for the power generation state variable, the binary indicator variable, and the local decision variable of the bus node, thereby obtaining the node solution value of the bus node.

3. The method according to claim 1, characterized in that, The step of generating the enhanced cut plane based on the node solution value includes: Starting from the second-to-last level, traverse each bus node of the scene path backward, refine the corresponding feasible interval based on the node solution value of the bus node, and determine the binary indicator variable of the refined feasible interval. Determine the parent node of the bus node, and construct a Lagrange dual model based on the state value of the parent node for the power generation state variable, the variable value for the binary indicator variable, and the approximate power generation cost of the bus node; Solving the Lagrange dual model yields the optimal dual multipliers and intercepts; The enhanced cut plane is generated based on the optimal dual multiplier and the intercept.

4. The method according to claim 3, characterized in that, Solving the Lagrange dual model to obtain the optimal dual multipliers and intercepts includes: The first cutting plane is generated according to the Pareto optimal Lagrange cutting plane strategy; Maximize the first cutting plane to solve the Lagrange dual model and obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model; The intercept is calculated based on the optimal dual multiplier and the model output value.

5. The method according to claim 3, characterized in that, Solving the Lagrange dual model to obtain the optimal dual multipliers and intercepts includes: The second cutting plane is generated based on the square minimization cutting plane strategy; The Lagrange dual model is solved by minimizing the second cutting plane to obtain the optimal model value; wherein, the optimal model value includes the optimal dual multiplier and the model output value of the Lagrange dual model; The intercept is calculated based on the optimal dual multiplier and the model output value.

6. The method according to any one of claims 1 to 5, characterized in that, The step of power dispatching the power network based on the node solution value, the target power generation state, and the target decision value includes: Calculate the upper bound of the feasible region based on the node solution values; The lower bound of the feasible region is calculated using the reference scheduling model based on the target power generation state and the target decision value; Calculate the gap between the lower bound of the feasible region and the upper bound of the feasible region; If the gap is less than or equal to a preset threshold, power dispatch is performed based on the target power generation state and the target decision value.

7. The method according to claim 6, characterized in that, The step of calculating the upper bound of the feasible region based on the node solution values ​​includes: Calculate the path cost of the scene path based on the node solution value; Calculate the mean and standard deviation of the path cost; The upper bound of the feasible region is calculated based on the mean and the standard deviation.

8. A power dispatching device, characterized in that, The device includes: The acquisition module is used to acquire the power dispatch model of the power network; wherein, the power dispatch model is a multi-stage stochastic mixed integer programming model defined on a scenario tree, the scenario tree includes bus nodes connected sequentially according to the dispatch stages, and the bus nodes have generation state variables and local decision variables; The first determining module is used to determine the feasible interval of the power generation state variable and the binary indicator variable of the feasible interval; wherein, the binary indicator variable is used to indicate whether the power generation state variable is located within the feasible interval; The second determining module is used to determine an approximate model of the power dispatching model based on the feasible interval and the binary indicator variable, so as to obtain a reference dispatching model; The sampling module is used to sample the scenario tree to obtain the scenario path, and solve the generation state variable, the binary indicator variable and the local decision variable of each bus node in the scenario path through the reference scheduling model to obtain the node solution value; The generation module is used to generate an enhanced cutting plane based on the node solution values; The calculation module is used to calculate the optimal solution of the power generation state variable of the target bus node to obtain the target power generation state based on the enhanced cut plane, and to calculate the optimal solution of the local decision variable of the target bus node to obtain the target decision value; wherein, the target bus node is the root node of the scene tree; The scheduling module is used to perform power scheduling on the power network based on the target power generation status and the target decision value.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 7.

Images